Local full-text search across your Google Drive. Indexes file metadata and document content into SQLite FTS5 for instant, private search.
Local full-text search across your Google Drive, built for the Vellum assistant.
Indexes file names, paths, and document content into a SQLite FTS5 database so the assistant can find files, snippets, and Drive links in milliseconds β without calling the Google Drive API for every lookup.
Google Drive folder
β crawl metadata + export supported docs
Indexing engine
β extract text, chunk, preserve metadata
SQLite FTS5 database (local)
β full-text search
Assistant tool results
β file names, paths, snippets, Drive links
All indexing stays on your machine. Document text never leaves your computer.
assistant plugins install google-drive-search
The plugin uses your assistant's managed Google connection β no API keys or tokens to handle. If Google isn't connected yet:
assistant oauth connect google
Approve the consent screen (Drive scope is included in the managed connection).
Ask the assistant:
"Index my Google Drive"
It runs drive_index, which crawls your Drive and builds the local search index. Large Drives can take a few minutes; the crawl keeps running even if the tool call times out β check progress with drive_status.
| Tool | Purpose |
|---|---|
drive_search | Search indexed Drive files by keyword. Returns names, paths, snippets, and Drive links. |
drive_status | Check index status: file count, size, last build metrics, connection status. |
drive_index | Build or update the index. Supports full rebuild and incremental updates. |
A config.json is created in the plugin storage directory on first run:
{
"rootFolderId": "root",
"maxDepth": -1,
"indexContentTypes": [
"application/vnd.google-apps.document",
"application/vnd.google-apps.spreadsheet",
"application/vnd.google-apps.presentation",
"text/plain",
"text/markdown"
],
"maxContentBytes": 10485760
}
rootFolderId: Drive folder ID to crawl. "root" = entire My Drive. Use a specific folder ID to index a subtree.maxDepth: Folder traversal depth. = unlimited, = only top-level files.-10maxContentBytes: Max file size to export content from. Files larger than this get metadata-only indexing.Content indexing (text extracted and searchable):
Metadata-only indexing (name and path searchable):
Apache-2.0